aboutsummaryrefslogtreecommitdiffstats
path: root/scalding/src/main/scala
diff options
context:
space:
mode:
authorBryan Newbold <bnewbold@archive.org>2018-08-20 18:43:11 -0700
committerBryan Newbold <bnewbold@archive.org>2018-08-21 21:25:56 -0700
commit29eb22bc60193e6f919d98ec14a39ecd53477331 (patch)
treefd3c99ce853bc5036dc2572eb04c1d3b1f6b8a38 /scalding/src/main/scala
parent39bf4b57cd552e8042bfa25565b390cb2a456ab0 (diff)
downloadsandcrawler-29eb22bc60193e6f919d98ec14a39ecd53477331.tar.gz
sandcrawler-29eb22bc60193e6f919d98ec14a39ecd53477331.zip
use grobid0:metadata, not tei_json
This is for efficiency. I had forgotten that the extract script actually writes this path!
Diffstat (limited to 'scalding/src/main/scala')
-rw-r--r--scalding/src/main/scala/sandcrawler/GrobidScorable.scala10
-rw-r--r--scalding/src/main/scala/sandcrawler/GrobidScorableDumpJob.scala10
2 files changed, 10 insertions, 10 deletions
diff --git a/scalding/src/main/scala/sandcrawler/GrobidScorable.scala b/scalding/src/main/scala/sandcrawler/GrobidScorable.scala
index d7a1eea..e510f75 100644
--- a/scalding/src/main/scala/sandcrawler/GrobidScorable.scala
+++ b/scalding/src/main/scala/sandcrawler/GrobidScorable.scala
@@ -24,10 +24,10 @@ class GrobidScorable extends Scorable with HBasePipeConversions {
getSource(args)
.read
// Can't just "fromBytesWritable" because we have multiple types?
- .toTypedPipe[(ImmutableBytesWritable,ImmutableBytesWritable,ImmutableBytesWritable)](new Fields("key", "tei_json", "status_code"))
- .filter { case (_, tei_json, status_code) => tei_json != null && status_code != null }
- .map { case (key, tei_json, status_code) =>
- (Bytes.toString(key.copyBytes()), Bytes.toString(tei_json.copyBytes()), Bytes.toLong(status_code.copyBytes()))
+ .toTypedPipe[(ImmutableBytesWritable,ImmutableBytesWritable,ImmutableBytesWritable)](new Fields("key", "metadata", "status_code"))
+ .filter { case (_, metadata, status_code) => metadata != null && status_code != null }
+ .map { case (key, metadata, status_code) =>
+ (Bytes.toString(key.copyBytes()), Bytes.toString(metadata.copyBytes()), Bytes.toLong(status_code.copyBytes()))
}
// TODO: Should I combine next two stages for efficiency?
.collect { case (key, json, StatusOK) => (key, json) }
@@ -37,7 +37,7 @@ class GrobidScorable extends Scorable with HBasePipeConversions {
object GrobidScorable {
def getHBaseSource(table : String, host : String) : HBaseSource = {
- HBaseBuilder.build(table, host, List("grobid0:tei_json", "grobid0:status_code"), SourceMode.SCAN_ALL)
+ HBaseBuilder.build(table, host, List("grobid0:metadata", "grobid0:status_code"), SourceMode.SCAN_ALL)
}
def jsonToMapFeatures(key : String, json : String) : MapFeatures = {
diff --git a/scalding/src/main/scala/sandcrawler/GrobidScorableDumpJob.scala b/scalding/src/main/scala/sandcrawler/GrobidScorableDumpJob.scala
index 5e06f9b..05e7074 100644
--- a/scalding/src/main/scala/sandcrawler/GrobidScorableDumpJob.scala
+++ b/scalding/src/main/scala/sandcrawler/GrobidScorableDumpJob.scala
@@ -26,13 +26,13 @@ class GrobidScorableDumpJob(args: Args) extends JobBase(args) {
val pipe = GrobidScorable.getHBaseSource(args("hbase-table"), args("zookeeper-hosts"))
.read
// Can't just "fromBytesWritable" because we have multiple types?
- .toTypedPipe[(ImmutableBytesWritable,ImmutableBytesWritable,ImmutableBytesWritable)](new Fields("key", "tei_json", "status_code"))
- .filter { case (_, tei_json, status_code) =>
+ .toTypedPipe[(ImmutableBytesWritable,ImmutableBytesWritable,ImmutableBytesWritable)](new Fields("key", "metadata", "status_code"))
+ .filter { case (_, metadata, status_code) =>
grobidHbaseRows.inc
- tei_json != null && status_code != null
+ metadata != null && status_code != null
}
- .map { case (key, tei_json, status_code) =>
- (Bytes.toString(key.copyBytes()), Bytes.toString(tei_json.copyBytes()), Bytes.toLong(status_code.copyBytes()))
+ .map { case (key, metadata, status_code) =>
+ (Bytes.toString(key.copyBytes()), Bytes.toString(metadata.copyBytes()), Bytes.toLong(status_code.copyBytes()))
}
// TODO: Should I combine next two stages for efficiency?
.collect { case (key, json, 200) =>